Gut Microbiome – Identifying Children with Type 1 Diabetes

Artificial intelligence fingerprints gut flora, linking clusters of microbiome species with type 1 diabetes development. 

The healthy gut flora contains hundreds of thousands of microorganisms, making up 150 times more genes than its human host. While a healthy and diverse gut microbial composition is essential for digestion, vitamin synthesis, and healthy immune response, dysbiosis of the gastrointestinal tract may alter vital metabolic processes. In recent years, it has been proposed that changes in the gut microbiome may even play a role in autoimmunity, an important mechanism involved in developing type 1 diabetes mellitus (T1DM).  

T1DM, previously known as juvenile diabetes or insulin-dependent diabetes, occurs when the pancreas makes little to no insulin due to the destruction of pancreatic β-cells by one’s immune system. It is usually diagnosed in childhood or adolescence, although rarely in adults. Genetic and environmental factors are known to contribute to the development of T1DM. However, the causative mechanisms are not entirely understood, and risk identification presents a challenge due to practical, diagnostic, and therapeutic implications. In a new study published in Endocrine Society’s Journal of Clinical Endocrinology and Metabolism, researchers analyzed fecal samples of pediatric patients to characterize the microbial fingerprint of people with T1DM and identify clusters of taxa that may be associated with altered metabolic pathways.  

Using a form of artificial intelligence called machine learning analysis, researchers analyzed fecal samples of 56 pediatric patients in Italy. All children were initially admitted to the emergency department and then sent to the Regional Center for Pediatric Diabetes inpatient pediatric clinic. Thirty-one children newly diagnosed with T1DM and 25 non-diabetic controls (healthy donors) were selected to participate in the study. Patients with gastrointestinal illnesses, a recent antibiotic or probiotic use, or other forms of diabetes were excluded. Using Next-Generation Sequencing, microbial DNA was extracted, amplified, and sequenced. The hypervariable regions of 16S ribosomal genes were used to determine each sample’s composition, relative abundance, and biodiversity.   

Two machine learning analyses (Random Forest and l1l2 algorithms), along with compositional and biodiversity analysis and statistical models, examined data regarding Phylum, Class, Order, Family, Genus, Species, relative abundance, and diversity. The samples identified 1606 Operational Taxonomic Units (OTUs) in the study group and 1552 OTUs in healthy donors. Upon analysis, researchers found significant differences in the gut microbiota composition of patients with T1DM compared to healthy subjects. Of most noticeable differences were higher relative abundances of B. stercoris, B. fragilis, B. intestinalis, B. bifidum, Gammaproteobacteria, Holdemania, and Synergistetes species and lower abundances of B. vulgatus, Deltaproteobacteria, Parasutterella, Lactobacillus, and Turicibacter species. In addition, healthy donors had a more diverse microbial flora than people with diabetes, although some analyses attributed this to differences in geographical locations. The Bacteroides genus (except for B. vulgatus) and Synergistetes subphylum were identified as the most significant organisms of interest in T1DM.  

Various metabolites produced by certain taxonomic groups have mechanisms that overlap with the pathology of metabolic diseases. The gene content analysis allowed the profiling of these metabolic pathways, linking them to their respective organisms. One-hundred and seven of the 712 metabolic pathways found in diabetic patients were associated with glucose metabolism. Fructose and mannose metabolism, starch and sucrose metabolism, pentose phosphate pathway, and galactose metabolism were the most common. Of note, reports have shown an increased risk of insulin resistance in patients with type 2 diabetes who had higher iron levels in tissues. Several of these iron metabolism pathways were also identified.  

In recent reports, CD8 T-cells, which are involved in autoimmunity, had been shown to cross-react with the epitopes of B. stercoris. Interestingly, B. stercoris was more abundant in T1DM patients than healthy donors. Studies have suggested that cross-reactivity of this bacterial strain can activate these cytotoxic lymphocytes leading to the destruction of pancreatic cells and the development of T1DM. However, the impact of the microbiota on local immune cell function is still unclear.  

In summary, the study showed a significant shift in the microbiota composition in children newly diagnosed with T1DM. Species in the Bacteroides genus and Synergistetes subphylum were significantly more abundant in the diabetic group. Also, higher relative abundances of gammaproteobacterial and Enterobacteriales species were observed – a finding has also been reported in patients with type 2 diabetes and was thought to be associated with impaired fasting blood glucose and intestinal permeability. While other research has reported data on two polymorphic regions of the 16S gene, this study included seven polymorphic regions adding to this investigation’s strengths. The variety of analyses and approaches used in this study also adds to the robustness and coherence of the results, indicating a need to research further the role of the gut microbiome in patients with diabetes.  

Practice Pearls: 

• Significant shifts in microbiota composition have been seen in children newly diagnosed with T1DM. 
• Species in the Bacteroides genus (except for B. vulgatus) and Synergistetes subphylum may be important indicators of metabolic disease. 
• CD8 T-cells, which are involved in autoimmunity, have been shown to cross-react with the epitopes of B. stercoris. 

References for “Gut Microbiome – Identifying Children with Type 1 Diabetes”:
Biassoni R, Di Marco E, Squillario M, et al. Gut Microbiota in T1DM-Onset Pediatric Patients: Machine-Learning Algorithms to Classify Microorganisms as Disease Linked. J Clin Endocrinol Metab. 2020;105(9):dgaa407. doi:10.1210/clinem/dgaa407 

Culina S, Lalanne AI, Afonso G, Cerosaletti K, et al. Islet-reactive CD8(+) T cell frequencies in the pancreas but not in blood, distinguish type 1 diabetic patients from healthy donors. Sci Immunol. 2018;3. PII:eaao4013. 

Melinda Rodriguez, PharmD Candidate 2021, Lake Erie College of Osteopathic Medicine – L|E|C|O|M School of Pharmacy